First functioning version of SW-only clip mask creator

This version passes the AAClips SampleApp slide test (i.e., single level boolean operations).

[bsalomon, 2012-05-16 15:36:10]

http://codereview.appspot.com/6208072/diff/1/src/gpu/GrClipMaskManager.cpp
File src/gpu/GrClipMaskManager.cpp (right):

http://codereview.appspot.com/6208072/diff/1/src/gpu/GrClipMaskManager.cpp#ne...
src/gpu/GrClipMaskManager.cpp:67: return clipIn.requiresAA();
We do draw some AA paths using GPU. Currently, just hairlines and convex paths.
It seems like we need a better mechanism here.

http://codereview.appspot.com/6208072/diff/1/src/gpu/GrSoftwarePathRenderer.cpp
File src/gpu/GrSoftwarePathRenderer.cpp (right):

http://codereview.appspot.com/6208072/diff/1/src/gpu/GrSoftwarePathRenderer.c...
src/gpu/GrSoftwarePathRenderer.cpp:167: bool GrSWMaskHelper::init(const GrIRect&
pathDevBounds,
Maybe init should just take a matrix as a required param instead of useMatrix
and translate?

[robertphillips, 2012-05-16 18:13:58]

http://codereview.appspot.com/6208072/diff/1/src/gpu/GrClipMaskManager.cpp
File src/gpu/GrClipMaskManager.cpp (right):

http://codereview.appspot.com/6208072/diff/1/src/gpu/GrClipMaskManager.cpp#ne...
src/gpu/GrClipMaskManager.cpp:67: return clipIn.requiresAA();
This new path demonstrates how I intend to limit the application of the SW-only
path. Right now it only fires on the complexclip_aa GM. The complexclip2* GMs
don't use the SW-only path b.c. they only use rects & round rects (all convex).

Note that the SW-only path is currently behind a compiler flag so won't be used
in production yet.

http://codereview.appspot.com/6208072/diff/1/src/gpu/GrSoftwarePathRenderer.cpp
File src/gpu/GrSoftwarePathRenderer.cpp (right):

http://codereview.appspot.com/6208072/diff/1/src/gpu/GrSoftwarePathRenderer.c...
src/gpu/GrSoftwarePathRenderer.cpp:167: bool GrSWMaskHelper::init(const GrIRect&
pathDevBounds,
I would like to hold off on this change for right now. I have a todo to figure
out why the SoftwarePathRender path requires it but the SW-only path does not.

[bsalomon, 2012-05-17 00:04:30]

LGTM.

On 2012/05/16 18:13:58, robertphillips wrote:
> http://codereview.appspot.com/6208072/diff/1/src/gpu/GrClipMaskManager.cpp
> File src/gpu/GrClipMaskManager.cpp (right):
> 
>
http://codereview.appspot.com/6208072/diff/1/src/gpu/GrClipMaskManager.cpp#ne...
> src/gpu/GrClipMaskManager.cpp:67: return clipIn.requiresAA();
> This new path demonstrates how I intend to limit the application of the
SW-only
> path. Right now it only fires on the complexclip_aa GM. The complexclip2* GMs
> don't use the SW-only path b.c. they only use rects & round rects (all
convex).
> 
> Note that the SW-only path is currently behind a compiler flag so won't be
used
> in production yet.
> 

Yup, OK. We can discuss the right mechanism when I'm back in the office. Maybe
we pass a flag down the PR chain telling it not to select the SW PR?

>
http://codereview.appspot.com/6208072/diff/1/src/gpu/GrSoftwarePathRenderer.cpp
> File src/gpu/GrSoftwarePathRenderer.cpp (right):
> 
>
http://codereview.appspot.com/6208072/diff/1/src/gpu/GrSoftwarePathRenderer.c...
> src/gpu/GrSoftwarePathRenderer.cpp:167: bool GrSWMaskHelper::init(const
GrIRect&
> pathDevBounds,
> I would like to hold off on this change for right now. I have a todo to figure
> out why the SoftwarePathRender path requires it but the SW-only path does not.

Ok by me. It'd be nice if the GrSWMaskRenderer didn't rely on the draw state of
the context at all.

[robertphillips, 2012-05-17 12:21:23]

committed as r3984